R help - Jun 2008 - ifelse and "&&" vs "&"

Hi,

I noticed whether some one could explain why "&" and
"&&" behave differently in data frame transformations.

Consider the following :

a<-data.frame(r=c(0,0,2,3),g=c(0,2,0,2.1))

Then:
> transform(a,R=ifelse(r>0 && g> 0,log(r/g),NA))
  r   g  R
1 0 0.0 NA
2 0 2.0 NA
3 2 0.0 NA
4 3 2.1 NA

but 
> transform(a,R=ifelse(r>0 & g> 0,log(r/g),NA))
  r   g         R
1 0 0.0        NA
2 0 2.0        NA
3 2 0.0        NA
4 3 2.1 0.3566749


If my understanding of the differences between "&" and
"&&" and how 'transform' works are accurate, both
statements should produce the same output.


I got the same behaviour in Windows XP Pro 32-bit (running R v 2.7) and Ubuntu
Hardy (running the same version of R).


Thanks

Christos Argyropoulos

University of Pittsburgh Medical Center
_________________________________________________________________
Discover the new Windows Vista

E

Hi there,

Simplifying your example a bit, we have:
> a=data.frame(r=c(0,0,1,1),g=c(0,1,0,1))
> transform(a,R=(r>0 && g>0))
  r g     R
1 0 0 FALSE
2 0 1 FALSE
3 1 0 FALSE
4 1 1 FALSE
> transform(a,R=(r>0 & g>0))
  r g     R
1 0 0 FALSE
2 0 1 FALSE
3 1 0 FALSE
4 1 1  TRUE
>
...but actually, in the && case the R column values are not calculated
from the corresponding values of r and g: the FALSE is effectively
calculated in row 1 and then copied down. Compare with this:
> a=data.frame(r=c(1,1,0,0),g=c(1,0,1,0))
> transform(a,R=(r>0 && g>0))
  r g    R
1 1 1 TRUE
2 1 0 TRUE
3 0 1 TRUE
4 0 0 TRUE
> transform(a,R=(r>0 & g>0))
  r g     R
1 1 1  TRUE
2 1 0 FALSE
3 0 1 FALSE
4 0 0 FALSE
>
This is actually explained in the ?!base man page: "& and &&
indicate
logical AND and | and || indicate logical OR. The shorter form performs
elementwise comparisons in much the same way as arithmetic operators. The
longer form evaluates left to right examining only the first element of
each vector. Evaluation proceeds only until the result is determined."

Hope this helps,

Toby Marthews


Le Mer 18 juin 2008 15:10, Christos Argyropoulos a ?crit
:
>
> Hi,
>
> I noticed whether some one could explain why "&" and
"&&" behave
> differently in data frame transformations.
>
> Consider the following :
>
> a<-data.frame(r=c(0,0,2,3),g=c(0,2,0,2.1))
>
> Then:
>
>> transform(a,R=ifelse(r>0 && g> 0,log(r/g),NA))
>
>   r   g  R
> 1 0 0.0 NA
> 2 0 2.0 NA
> 3 2 0.0 NA
> 4 3 2.1 NA
>
> but
>
>> transform(a,R=ifelse(r>0 & g> 0,log(r/g),NA))
>   r   g         R
> 1 0 0.0        NA
> 2 0 2.0        NA
> 3 2 0.0        NA
> 4 3 2.1 0.3566749
>
>
> If my understanding of the differences between "&" and
"&&" and how
> 'transform' works are accurate, both statements should produce the
same
> output.
>
>
> I got the same behaviour in Windows XP Pro 32-bit (running R v 2.7) and
> Ubuntu Hardy (running the same version of R).
>
>
> Thanks
>
> Christos Argyropoulos
>
> University of Pittsburgh Medical Center

On Wed, Jun 18, 2008 at 3:10 PM, Christos Argyropoulos
<argchris at hotmail.com> wrote:
>
> Hi,
>
> I noticed whether some one could explain why "&" and
"&&" behave differently in data frame transformations.
>
> Consider the following :
>
> a<-data.frame(r=c(0,0,2,3),g=c(0,2,0,2.1))
>
> Then:
>
>> transform(a,R=ifelse(r>0 && g> 0,log(r/g),NA))
>
>  r   g  R
> 1 0 0.0 NA
> 2 0 2.0 NA
> 3 2 0.0 NA
> 4 3 2.1 NA
>
> but
>
>> transform(a,R=ifelse(r>0 & g> 0,log(r/g),NA))
>  r   g         R
> 1 0 0.0        NA
> 2 0 2.0        NA
> 3 2 0.0        NA
> 4 3 2.1 0.3566749
>
>
> If my understanding of the differences between "&" and
"&&" and how 'transform' works are accurate, both
statements should produce the same output.
>
>
> I got the same behaviour in Windows XP Pro 32-bit (running R v 2.7) and
Ubuntu Hardy (running the same version of R).
>
>
> Thanks
>
> Christos Argyropoulos
>
> University of Pittsburgh Medical Center
> _________________________________________________________________

from ?"&" : " The shorter form performs elementwise
comparisons in
much the same way as arithmetic operators. The longer form evaluates
left to right examining only the first element of each vector. "

Thus,
> a$r & a$g
[1] FALSE FALSE FALSE  TRUE
> a$r && a$g
[1] FALSE

ifelse takes a vector as argument. isince && only gives a single
value,  ifelse(r>0 && g> 0,log(r/g),NA) will only return NA, which
then is recycled by transform. When using &, ifelse returns a vector,
and this vector is appended to the data frame.

/Gustaf

-- 
Gustaf Rydevik, M.Sci.
tel: +46(0)703 051 451
address:Essingetorget 40,112 66 Stockholm, SE
skype:gustaf_rydevik